<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: nilsherzig</title><link>https://news.ycombinator.com/user?id=nilsherzig</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 07 Sep 2026 12:25:48 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=nilsherzig" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by nilsherzig in "Signing TLS handshakes inside a TPM"]]></title><description><![CDATA[
<p>Still flags as 100% LLM written <a href="https://www.pangram.com/history/cc9b6131-c772-4b3b-9a5d-8e81850a852f?ucc=QjpX0hxYnaS" rel="nofollow">https://www.pangram.com/history/cc9b6131-c772-4b3b-9a5d-8e81...</a></p>
]]></description><pubDate>Mon, 07 Sep 2026 07:57:31 +0000</pubDate><link>https://news.ycombinator.com/item?id=49595303</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49595303</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49595303</guid></item><item><title><![CDATA[New comment by nilsherzig in "The revolt of the reader"]]></title><description><![CDATA[
<p>update: <a href="https://news.ycombinator.com/item?id=49582725">https://news.ycombinator.com/item?id=49582725</a> shared another page which already does the parsing and pangram requests. I wrote a little wrapper and userscript based extension around it (<a href="https://github.com/nilsherzig/hnslop" rel="nofollow">https://github.com/nilsherzig/hnslop</a>), hope that helps!</p>
]]></description><pubDate>Sun, 06 Sep 2026 12:28:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=49585886</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49585886</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49585886</guid></item><item><title><![CDATA[New comment by nilsherzig in "The revolt of the reader"]]></title><description><![CDATA[
<p>created a little wrapper and userscript based browser extension <a href="https://hnslop.nilsherzig.com/" rel="nofollow">https://hnslop.nilsherzig.com/</a></p>
]]></description><pubDate>Sun, 06 Sep 2026 12:26:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=49585875</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49585875</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49585875</guid></item><item><title><![CDATA[New comment by nilsherzig in "The revolt of the reader"]]></title><description><![CDATA[
<p>I have (an API, not an extension), but pangram is way to expensive for me to run, current (v4) pricing is $0.05 per 100 words. Should be absolutely doable if we split the cost between users tho.<p>I tried searching for good opensource / reasonably priced alternatives, pangram themselves even have some of their older architecture and training data on github/hugging face, but i never got that working reliably enough.</p>
]]></description><pubDate>Sun, 06 Sep 2026 09:23:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=49584678</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49584678</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49584678</guid></item><item><title><![CDATA[New comment by nilsherzig in "The OG Creator of Task Manager on Windows Built a New Task Manager"]]></title><description><![CDATA[
<p>checks out:<p>> In 2006, SoftwareOnline was sued by The Washington State Attorney General's Office for alleged violations of the Consumer Protection Act after complaints were made about two products called "Registry Cleaner" and "InternetShield".<p><a href="https://en.wikipedia.org/wiki/Dave_Plummer#:~:text=16%5D-,In,InternetShield%22%2E" rel="nofollow">https://en.wikipedia.org/wiki/Dave_Plummer#:~:text=16%5D-,In...</a></p>
]]></description><pubDate>Fri, 28 Aug 2026 20:21:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=49483734</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49483734</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49483734</guid></item><item><title><![CDATA[New comment by nilsherzig in "Ask HN: What is one simple thing LLMs are insanely bad at?"]]></title><description><![CDATA[
<p>Removing stuff without mentioning the removal.<p>Asking an LLM to remove an Idea from a Document often times results in an edit which explicitly states that this Idea is not relevant, instead of just removing all references to that Idea.<p>That might be useful for some form of "evolving" Documentation (so future readers know that this part of the search space was covered and deemed irrelevant) but is just overly verbose and confusing to read in most situations.</p>
]]></description><pubDate>Wed, 26 Aug 2026 07:18:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=49445167</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49445167</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49445167</guid></item><item><title><![CDATA[New comment by nilsherzig in "Ask HN: What is one simple thing LLMs are insanely bad at?"]]></title><description><![CDATA[
<p>You can get better results from less aligned models like Kimi K3. Still not actually funny, but at least it’s able to produce some unhinged stuff and I guess shock and twists are kinda related to humor?<p>Still missing the human connection of cause, so im not sure if this is a technical / skill issue in the first place.</p>
]]></description><pubDate>Wed, 26 Aug 2026 07:14:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=49445137</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49445137</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49445137</guid></item><item><title><![CDATA[New comment by nilsherzig in "Ask HN: What is one simple thing LLMs are insanely bad at?"]]></title><description><![CDATA[
<p>That's a Claude thing btw. Try using a harness which does not inject half a novel of instructions in combination with a different model. I would recommend Pi + GPT 5.6 Luna for a very capable and cheap test.<p>After using Claude (paid by work) for a couple of months, I was amazed how well instruction following works in other setups.</p>
]]></description><pubDate>Wed, 26 Aug 2026 07:09:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=49445103</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49445103</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49445103</guid></item><item><title><![CDATA[New comment by nilsherzig in "Letter from Dolly Parton to Eminem"]]></title><description><![CDATA[
<p>oh looks like they killed xcancel :/<p><a href="https://xcancel.com/Eminem/status/2092367272891310482" rel="nofollow">https://xcancel.com/Eminem/status/2092367272891310482</a></p>
]]></description><pubDate>Wed, 26 Aug 2026 07:01:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=49445038</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49445038</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49445038</guid></item><item><title><![CDATA[New comment by nilsherzig in "AI-Generated Images Discourage Me from Reading Your Blog"]]></title><description><![CDATA[
<p>I don't see a lot of value in that, but one could supply a lot of semi structured extra information which could aid the readers LLM to answer questions the reader might have with more accuracy / factual grounding.</p>
]]></description><pubDate>Wed, 05 Aug 2026 07:20:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=49179606</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=49179606</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49179606</guid></item><item><title><![CDATA[New comment by nilsherzig in "I built a mmWave material classification radar"]]></title><description><![CDATA[
<p>love the background music in combination with the flying fishes wallpaper in the first video haha<p>very cool project</p>
]]></description><pubDate>Tue, 30 Jun 2026 19:09:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=48737787</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=48737787</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48737787</guid></item><item><title><![CDATA[New comment by nilsherzig in "Pledging another $400k to the Zig software foundation"]]></title><description><![CDATA[
<p>If you're unsure about spending the time to learn Zig, I really recommend watching the following interview with the creator of Zig <a href="https://www.youtube.com/watch?v=iqddnwKF8HQ" rel="nofollow">https://www.youtube.com/watch?v=iqddnwKF8HQ</a> convinced me more than any design doc or blogpost could</p>
]]></description><pubDate>Mon, 22 Jun 2026 17:52:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=48633488</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=48633488</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48633488</guid></item><item><title><![CDATA[New comment by nilsherzig in "Surprising economics of load-balanced systems"]]></title><description><![CDATA[
<p>Why would anyone think that it would get linearly worse? What's the (wrong) assumption there?</p>
]]></description><pubDate>Fri, 19 Jun 2026 22:56:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=48604270</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=48604270</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48604270</guid></item><item><title><![CDATA[New comment by nilsherzig in "A new bill takes aim at government pressure to silence lawful online speech"]]></title><description><![CDATA[
<p>I like MAGIC CARPET "Maritime Augmented Guidance with Integrated Controls for Carrier Approach and Recovery Precision Enabling Technologies"</p>
]]></description><pubDate>Fri, 19 Jun 2026 21:12:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=48603294</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=48603294</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48603294</guid></item><item><title><![CDATA[New comment by nilsherzig in "A key remapping daemon for Linux"]]></title><description><![CDATA[
<p>Great software. Especially the "lettermod" feature. It allows you to configure a key to switch to a layer / secondary actions if held for some time, but only if no other key was pressed within some time before. So secondary key actions don't happen while the user is actively typing a text. This makes it very easy to add homerow mods (adding Ctrl, Alt, etc to the homerow keys) without the struggle of accidently triggering them.</p>
]]></description><pubDate>Sat, 13 Jun 2026 08:18:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=48514823</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=48514823</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48514823</guid></item><item><title><![CDATA[New comment by nilsherzig in "My smart sleep mask broadcasts users' brainwaves to an open MQTT broker"]]></title><description><![CDATA[
<p>yes but <a href="https://aimilios.bearblog.dev/reverse-engineering-sleep-mask/#:~:text=It%20connected,occupancy" rel="nofollow">https://aimilios.bearblog.dev/reverse-engineering-sleep-mask...</a> reads like they connected to MQTT and received data from IOT devices on there, not using BLE</p>
]]></description><pubDate>Tue, 17 Feb 2026 14:23:44 +0000</pubDate><link>https://news.ycombinator.com/item?id=47047793</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=47047793</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47047793</guid></item><item><title><![CDATA[New comment by nilsherzig in "My smart sleep mask broadcasts users' brainwaves to an open MQTT broker"]]></title><description><![CDATA[
<p>Can someone explain the other iot devices using the same broker? I tried cross referencing the feature list, information about the user base, kickstarter origin and flutter app with some search results and I’m pretty sure that I found the company and product in question. But they don’t (publicly) produce iot devices? Sooo I’m wondering if different companies are streaming their data into a shared sink and why they would do that?</p>
]]></description><pubDate>Sun, 15 Feb 2026 07:48:21 +0000</pubDate><link>https://news.ycombinator.com/item?id=47021827</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=47021827</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47021827</guid></item><item><title><![CDATA[New comment by nilsherzig in "Map showing most notable people per region"]]></title><description><![CDATA[
<p>Not mine. Was already posted on here, but the last time it got traction was a couple of years back.</p>
]]></description><pubDate>Tue, 10 Feb 2026 16:53:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=46962775</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=46962775</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46962775</guid></item><item><title><![CDATA[Map showing most notable people per region]]></title><description><![CDATA[
<p>Article URL: <a href="https://tjukanovt.github.io/notable-people">https://tjukanovt.github.io/notable-people</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46962774">https://news.ycombinator.com/item?id=46962774</a></p>
<p>Points: 2</p>
<p># Comments: 2</p>
]]></description><pubDate>Tue, 10 Feb 2026 16:53:30 +0000</pubDate><link>https://tjukanovt.github.io/notable-people</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=46962774</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46962774</guid></item><item><title><![CDATA[New comment by nilsherzig in "AI doesn’t reduce work, it intensifies it"]]></title><description><![CDATA[
<p>I think the randomness is addicting. While writing a prompt often doesn't result in the perfect outcome, it very well could. Pressing the "prompt lever" (again and again), waiting for the result to show up looks a lot like gambling.<p>See "Variable Ratio Schedule" <a href="https://www.simplypsychology.org/schedules-of-reinforcement.html#Fixed-Ratio-Schedule:~:text=This%20type,reinforcement,-%2E" rel="nofollow">https://www.simplypsychology.org/schedules-of-reinforcement....</a></p>
]]></description><pubDate>Tue, 10 Feb 2026 14:40:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=46960270</link><dc:creator>nilsherzig</dc:creator><comments>https://news.ycombinator.com/item?id=46960270</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46960270</guid></item></channel></rss>